Vol 440|20 April 2006|doi:10.1038/nature04656

LETTERS Enhanced bacterial clearance and sepsis resistance in caspase-12-deficient mice Maya Saleh1†, John C. Mathison2, Melissa K. Wolinski1, Steve J. Bensinger1, Patrick Fitzgerald1, Nathalie Droin1, Richard J. Ulevitch2, Douglas R. Green1*† & Donald W. Nicholson*3

Caspases function in both apoptosis and inflammatory cytokine processing and thereby have a role in resistance to sepsis1. Here we describe a novel role for a caspase in dampening responses to bacterial infection. We show that in mice, gene-targeted deletion of caspase-12 renders animals resistant to peritonitis and septic shock. The resulting survival advantage was conferred by the ability of the caspase-12-deficient mice to clear bacterial infection more efficiently than wild-type littermates. Caspase-12 dampened the production of the pro-inflammatory cytokines interleukin (IL)-1b, IL-18 (interferon (IFN)-g inducing factor) and IFN-g, but not tumour-necrosis factor-a and IL-6, in response to various bacterial components that stimulate Toll-like receptor and NOD pathways. The IFN-g pathway was crucial in mediating survival of septic caspase-12-deficient mice, because administration of neutralizing antibodies to IFN-g receptors ablated the survival advantage that otherwise occurred in these animals. Mechanistically, caspase-12 associated with caspase-1 and inhibited its activity. Notably, the protease function of caspase-12 was not necessary for this effect, as the catalytically inactive caspase-12 mutant Cys299Ala also inhibited caspase-1 and IL-1b production to the same extent as wild-type caspase-12. In this regard, caspase12 seems to be the cFLIP counterpart for regulating the inflammatory branch of the caspase cascade. In mice, caspase-12 deficiency confers resistance to sepsis and its presence exerts a dominant-negative suppressive effect on caspase-1, resulting in enhanced vulnerability to bacterial infection and septic mortality. To address the function of caspase-12 in apoptosis and immunity, caspase-12-null mutant C57Bl/6 mice were generated in which a neo/LacZ cassette replaced a segment of exon 2 of the caspase-12 gene (Casp12), shifting the downstream sequence out of the reading frame (see Methods and Supplementary Fig. 1a, b). Caspase-12 transcripts and protein were not detected in mutant mice by either polymerase chain reaction (not shown) or western blotting (Supplementary Fig. 1c). Deletion of Casp12 had no effect on the expression of neighbouring caspase genes in the inflammatory gene cluster—we observed comparable levels of lipopolysaccharide (LPS)-inducible caspase-11 transcripts and protein in the peripheral blood of wildtype and caspase-12-deficient mice (Supplementary Fig. 1d, e), and the expression of caspase-1 was similarly not altered by the deletion of Casp12 (Supplementary Fig. 1f). As found for mice deficient in other inflammatory caspases (Casp1 2/2 (refs 2, 3) and Casp11 2/2 mice4), caspase-12-deficient (Casp12 2/2) mice were indistinguishable in phenotype compared to age- and gender-matched wild-type littermates (physical examination, necropsy, histological evaluation, blood counts and chemistry, and behavioural and fertility tests; data not shown).

The association of human caspase-12 (the caspase-12 long variant, CASP12L) with susceptibility to sepsis in individuals of African descent5 led us to examine the response of the Casp12 þ/þ and Casp12 2/2 mice in a murine model of polymicrobial sepsis. To this end, we applied the colon ascendens stent peritonitis (CASP)6 model that closely mimics human sepsis. Whereas most of the wildtype mice succumbed to sepsis in the first 48 h after the peritonitis surgery, 60% of the Casp12 2/2 mice were resistant to sepsis and survived (log rank test, P ¼ 0.00065) (Fig. 1a), indicating that the presence of caspase-12 predisposed the mice to sepsis and that its absence conferred protection. Surprisingly, the Casp12 2/2 mice were sensitive to LPS overdose: 90% of the knockout mice died by 72 h after LPS injection (not shown). This finding contrasts with the resistance to LPS overdose that was previously reported for Casp1 2/2 (refs 2, 3) and Casp11 2/2 mice4, indicating that caspase-12 function in vivo is not redundant with that of these two related caspases. To dissect the mechanism responsible for the survival advantage of the caspase-12-deficient mice in this model, we measured blood bacterial content 12 h after CASP surgery. Casp12 2/2 mice showed a markedly lower number of bacterial colony-forming units (c.f.u.) per ml blood compared to wild-type littermates (Fig. 1b, c), suggesting that more efficient bacterial clearance occurs in the absence of caspase-12. We evaluated the number and type of commensal bacteria from both mouse genotypes, and did not find any significant difference between the microflora of the Casp12 2/2 compared to the wild-type mice (not shown). Therefore, the effect that we observed is most probably due to a difference in bacterial clearance. To determine whether this effect was restricted to abdominal infections or was a general characteristic of the caspase-12-deficient mice, a systemic route of infection was examined. Casp12 þ/þ and Casp12 2/2 mice were challenged intravenously with a defined dose of the Grampositive bacterium Listeria monocytogenes. Pathogen load was assessed by measuring c.f.u. in the spleen, liver and peripheral blood at varying time points after infection. Similar to what we observed in the CASP model, Casp12 2/2 mice cleared L. monocytogenes more efficiently than wild-type mice (Fig. 1d, e, and data not shown). Taken together, these results suggest that caspase-12 has a critical role in regulating the clearance of bacterial pathogens such that in the absence of caspase-12, both systemic and abdominal infections are better contained and resolved. We next sought to probe the signal transduction pathways and cytokine responses that might be affected by murine caspase-12. We chose to examine this in ex vivo splenocytes because we observed that the lacZ gene that was introduced into the caspase-12 targeted locus was highly induced by bacterial infection in the spleen (Fig. 2a). Splenocytes from Casp12 þ/þ and Casp12 2/2 mice were isolated and

1 Department of Cellular Immunology, La Jolla Institute for Allergy and Immunology, San Diego, California 92121, USA. 2Department of Immunology, The Scripps Research Institute, San Diego, California 92037, USA. 3Merck Research Laboratories, Rahway, New Jersey 07065-0900, USA. †Present addresses: Department of Medicine, McGill University, Montreal H3A 1A1, Canada (M.S.); Department of Immunology, St Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA (D.R.G.). *These authors contributed equally to this work.

1064

© 2006 Nature Publishing Group

LETTERS

NATURE|Vol 440|20 April 2006

analysed for their cytokine response to various Toll-like receptor (TLRs) ligands and the NOD2 ligand muramyl dipeptide (MDP). Notably, in response to some of the TLR/NOD ligands tested the caspase-12-deficient splenocytes produced more of the proinflammatory cytokines IL-1b, IL-18 (also known as IGIF) and IFN-g compared to wild-type cells (Fig. 2b), but not tumournecrosis factor (TNF)-a or IL-6 (Supplementary Fig. 2). As a control, we examined the relative expression of pro-IL-1b transcripts from Casp12 þ/þ and Casp12 2/2 macrophages in response to TLR ligands to determine whether the negative effect of caspase-12 on IL-1b production is due to inhibition of pro-IL-1b transcription. Our results indicate that the levels of pro-IL-1b message were not significantly different between the two mouse genotypes (Fig. 2c). The lack of effect of caspase-12 status on TNF and IL-6 production was in contrast to what was observed in humans with CASP12L, the expression of which dampened the levels of TNF as well as that of granulocyte–macrophage colony stimulating factor (GM-CSF) and

Figure 1 | Casp12 2/2 mice are resistant to sepsis and clear pathogens more efficiently than Casp12 1/1 mice. a, Sepsis was induced using the CASP model of polymicrobial infection. Survival was recorded every 3 h. Whereas more than 80% of Casp12 þ/þ mice succumbed to septic shock in the first 48 h after CASP, most Casp12 2/2 mice survived. b, Blood bacterial load in sepsis. Peripheral blood was collected 12 h after CASP, and serial dilutions were plated on blood agar plates, showing that Casp12 2/2 mice cleared the polymicrobial infection more efficiently than wild-type littermates. Each plate represents a different mouse (n ¼ 8 for both Casp12 þ/þ and Casp12 2/2). c, Quantification of c.f.u. per ml blood from the experiment described in b. Error bars indicate standard deviations. d, e, Listeria monocytogenes pathogen clearance. Casp12 þ/þ and Casp12 2/2 mice were challenged intravenously with a defined dose of L. monocytogenes, and pathogen clearance was determined by assessment of c.f.u. ml21 of bacteria in spleen (d) and liver (e). NS, not significant. n ¼ 3 per genotype per time point, indicated by individual symbols of the same colour.

IL-8 (ref. 5). It is therefore possible that murine caspase-12 and the human CASP12L variant may use non-overlapping as well as common signal transduction pathways in sepsis to regulate the inflammatory response. For example, whereas human CASP12L blocked NF-kB signalling in response to TNF or LPS, rodent caspase-12 had no effect on NF-kB regulation (our previous results5 and Supplementary Fig. 3). In both humans and mice, IFN-g is an established survival factor in sepsis6–8. Because the Casp12 2/2 mice produced noticeably more IFN-g in response to bacteria than wild-type mice, we tested whether this hyper-production of IFN-g might account for the protection of the knockout mice from septic shock. Neutralizing antibodies to murine IFN-g receptors were injected intraperitoneally in the Casp12 2/2 mice 1 h before initiation of sepsis by CASP and survival was assessed as described above. Disruption of IFN-g signalling ablated the survival advantage of caspase-12-deficient mice after CASP (Fig. 2d), indicating that caspase-12 negatively regulates or is dependent upon this pathway in vivo. We next examined whether the catalytic function of rodent caspase-12 was required for its effects on inflammation. For this purpose, the human monocytic THP-1 cell-line—that does not produce human caspase-12 (our previous results5 and data not shown)—was transfected with rodent caspase-12 or the catalytically inactive form, caspase-12 Cys299Ala (along with green fluorescent protein (GFP) as a transfection marker), and cytokine production was examined in the sorted GFP-positive cells. Caspase-12 specifically inhibited the production of IL-1b in response to TLR ligands (Fig. 3a) but did not have any effect on that of GM-CSF (Fig. 3b). Notably, mutation of the catalytic cysteine in caspase-12 did not abolish its inhibitory role on IL-1b production, indicating that the enzymatic function of caspase-12 is not required for this effect. Addition of zVAD-fmk to block caspase-1 in the cultures resulted in undetectable levels of IL-1b and IL-18 in the culture media but did not affect GM-CSF levels (data not shown). Therefore, in a manner similar to the way that cFLIP is a dominant-negative modulator of the ‘extrinsic’ cell death pathway9,10, caspase-12 seems to be a counterpart in the inflammatory caspase cascade. This is supported by other observations, namely (1) the hyper-production of IL-1b and IL-18 by Casp-12 2/2 splenocytes (Fig. 2b), (2) the inhibition of IL-1b production by overexpressed rodent caspase-12 in THP-1 cells (Fig. 3a) (with the absence of an effect on pro-IL-1b transcription (Fig. 2c)), and (3) the confinement of caspase-12 effects to caspase-1 substrates (for example, IL-1b and IL-18 (substrates of caspase-1) but not GM-CSF). Because caspase-12 seems to be a dominant-negative regulator of caspase-1, we tested this directly using two approaches. First, we examined the effect of rodent caspase-12 overexpression on the activity of caspase-1 in HEK 293T cells. Both wild-type caspase-12 and the catalytically inactive mutant caspase-12 Cys299Ala blocked the activity of caspase-1 on its preferred substrate Trp-Glu-His-Asp7-amino-4-trifluoromethyl coumarin (WEHD-AFC) (Fig. 4a). Second, we used an established system in which caspase-1 is spontaneously activated in a cell-free system11, in the presence of either a control protein or increasing concentrations of bacterially expressed and purified recombinant caspase-12, and examined the maturation of IL-1b by western blot analysis (Fig. 4b). Caspase-12 inhibited the production of the 17-kDa mature form of IL-1b in a dose-dependent manner, confirming that the dominant-negative inhibitory action of caspase-12 affected both synthetic and macromolecular substrates of caspase-1. To determine whether the inhibitory role of caspase-12 on caspase-1 is mediated through an association between the two proteins, a co-immunoprecipitation experiment was performed in HEK 293T cells, revealing that caspase-12 co-immunoprecipitates with caspase1. To control for specificity, we also examined the co-immunoprecipitation of caspase-12 with caspase-5 and caspase-9, both of which contain a CARD domain. Caspase-12 co-immunoprecipitated

© 2006 Nature Publishing Group

1065

LETTERS

NATURE|Vol 440|20 April 2006

Figure 2 | Caspase-12 dampens production of cytokines essential for sepsis survival. a, Caspase-12 is expressed in intestinal epithelial cells and is induced in the spleen in response to bacteria. Caspase-12 expression was examined by staining tissue sections from Casp12 2/2 mice that either underwent CASP (panels IV–VII) or not (panels II–III) for b-galactosidase expression (driven from the LacZ-neo cassette in the Casp12 targeted allele). Panel I: intestine section from a wild-type mouse was stained for b-galactosidase as a negative control. b, Enhanced production of IL-1b, IL-18 and IFN-g by Casp12 2/2 splenocytes. Splenocytes from Casp12 þ/þ (n ¼ 4) and Casp12 2/2 (n ¼ 4) mice were treated ex vivo with different TLR/NOD ligands for 24 h, and cytokine levels were measured from the

culture media, using bead-based multiplex immunoassays. Results are from two independent experiments. Error bars indicate standard deviations. c, Thioglycollate-elicited mouse peritoneal macrophages from Casp12 þ/þ (n ¼ 4) and Casp12 2/2 (n ¼ 4) mice were treated with TLR ligands, mRNA extracted and pro-IL-1b levels measured by quantitative real-time PCR. Individual animals are represented by individual symbols of the same colour. d, Inhibition of IFN-g signalling ablates the survival advantage of caspase-12-deficient mice after CASP. One hour before CASP, mice were injected intraperitoneally with 250 mg per mouse of neutralizing anti-mouse IFN-g receptor antibodies or isotype control (not shown). Survival to sepsis was assessed as in Fig. 1a.

to a lesser extent with caspase-5 (compared with caspase-1) and did not co-immunoprecipitate with caspase-9 (Fig. 4c). Whereas both caspase-1- and caspase-11-deficient mice are impaired in their ability to produce IL-1b and are resistant to lethal doses of endotoxin/LPS2,4, caspase-1-deficient mice are two- to threefold more susceptible to lethal Escherichia coli infection than wild-type mice12. Because caspase-12 is an inhibitor of caspase-1, the resistance to sepsis of Casp12 2/2 mice is probably associated with an initial hyper-production of these cytokines mediated by a derepression of caspase-1. The resulting beneficial effect of cytokine hyper-production is contrary to current dogma in sepsis research in which the initial cytokine ‘storm’ is viewed as harmful and has been the basis of many (albeit unsuccessful) clinical studies. Thus, in wildtype cells murine caspase-12 seems to attenuate the activity of caspase-1 that is normally essential for bacterial clearance and sepsis survival. To evaluate the role of caspase-12 in the haematopoietic system and understand whether its expression in other tissues (for example, intestine) is also required for its effects on pathogen clearance, bone marrow chimaera studies were performed. Casp12 þ/þ or Casp12 2/2 bone marrow was transplanted, respectively, into Casp12 2/2 or Casp12 þ/þ recipient mice. The ability of the chimaeric mice to clear polymicrobial infection was determined as above in response to CASP-induced sepsis. Whereas Casp12 þ/þ bone marrow reduced the ability of Casp12 2/2 mice to clear bacteria, the transfer of Casp12 2/2 bone marrow into Casp12 þ/þ host mice only partially improved their clearance ability and did not rescue them from sepsis (Supplementary Fig. 4 and data not shown). These results suggest that caspase-12 functions not only in the haematopoietic system, but may also modulate the inflammatory response in other tissues. It was recently reported that NOD2 modulates TLR2 signalling13,14. We therefore examined whether caspase-12 has a role in

this modulation. Splenocytes from wild-type and Casp12 2/2 mice were treated with the TLR2 ligand PAMcsk4 alone or in combination with the NOD2 ligand MDP, and the levels of IFN-g and IL-6 were measured. The inhibitory effect of NOD2 signalling on IFN-g production was not altered by the presence or absence of caspase12 (ref. 12 and Supplementary Fig. 5a), indicating that caspase-12 does not cooperate with NOD2 to regulate IFN-g levels. Similarly, the synergy between the TLR2 and NOD2 ligands in IL-6 production14 was not affected by the absence of caspase-12 (Supplementary Fig. 5b), suggesting that caspase-12 functions independently of NOD2 downstream of TLR signalling. Murine caspase-12 was previously proposed as a key mediator of endoplasmic reticulum (ER) stress-induced apoptosis15. Because apoptosis contributes adversely to sepsis progression1, we sought to evaluate the role of caspase-12 in ER stress-induced apoptosis in our system. We examined the response of primary mouse embryonic fibroblasts (MEFs)—isolated from wild-type and Casp12 2/2 mice— to apoptosis induced by the ER stressors brefeldin A, tunicamycin, thapsigargin and the calcium ionophore A23187. Casp12 þ/þ and Casp12 2/2 MEFs showed similar levels of cell death after treatment with these stressors (Supplementary Fig. 6). Similarly, the presence or absence of caspase-12 had no effect on apoptotic sensitivity to diverse stimuli, including activators of the extrinsic and intrinsic cell death pathways (data not shown). These data are consistent with those described for human caspase-12 in which no differential apoptosis sensitivity was observed5, as well as with other results16,17; however, they differ from what was reported in another study15. Because the mice described here were generated using a different targeting strategy than that of the mice described in the latter study15, the difference in results may relate to a difference in caspase-11 expression levels in those animals versus those we used. Caspase-11 is not only an important player in responses to endotoxin4, but it has

1066

© 2006 Nature Publishing Group

LETTERS

NATURE|Vol 440|20 April 2006

Figure 3 | The catalytic function of caspase-12 is not required for its inhibitory effect on IL-1b production. a, b, THP-1 cells (deficient in human caspase-12 expression) were reconstituted to express GFP in combination with vector control, rat caspase-12 or the catalytically inactive mutant Cys299Ala. Sorted live GFP-positive cells were either left undifferentiated or were differentiated with PMA before treatment with TLR ligands. Levels of IL-1b (a) and GM-CSF (b) were measured from the culture media using bead-based human immunoassays. Treatments were done in triplicate. Results are from two independent experiments. Error bars indicate standard deviations.

also been suggested to have a role in ER stress apoptosis18. It is therefore possible that caspase-11 rather than caspase-12 might have a role in ER stress-induced apoptosis in some settings. Our study describes a fundamentally new role for a caspase: dampening of responses to bacterial infection. Murine caspase-12 deficiency confers resistance to sepsis and its presence exerts a direct suppressive effect on caspase-1, resulting in enhanced vulnerability to bacterial infection and septic shock. Whereas the presence of caspase-12 seems to be detrimental for the individual during sepsis, in both mice (this study) and humans5, it remains to be determined whether murine caspase-12 and the human CASP12L variant modulate inflammation and bacterial clearance by overlapping or mutually exclusive mechanisms. Regardless, caspase-12 is detrimental to in vivo handling of systemic bacterial infections and predisposes to sepsis, thereby making it a potentially important target for future therapeutic strategies. METHODS Detailed methods and methods not described here can be found in Supplementary Information. Generation of Casp12 knockout mice and primary cell cultures. Casp12 knockout mice were generated by Deltagen under a consortium agreement involving Merck & Co. A targeting construct was engineered to disrupt the Casp12 gene in ES cells by homologous recombination. A segment of exon 2 was targeted for replacement by a cassette consisting of the b-galactosidase gene and the neomycin (neo) resistance gene. The Casp12 2/2 mice used in this study were backcrossed to homogeneity to a C57BL/6J background. MEFs were prepared from embryos harvested at day 14 of gestation. Mouse intestinal epithelial cells (IECs) were purified on discontinuous Percoll gradients from the small intestine. Induction of sepsis using the CASP procedure. Sepsis was induced in mice aged between 6–12 weeks by pressing caecal content into the abdomen through a stent

Figure 4 | Caspase-12 as well as the catalytically inactive mutant Cys299Ala associate with caspase-1 and block its activity. a, Caspase-1 substrate WEHD-AFC cleavage by HEK 293T lysates expressing either vector control or caspase-1 in combination with caspase-12 or caspase-12 Cys299Ala. Error bars indicate standard deviations. AFU, arbitrary fluorescent units. b, Caspase-12 inhibits inflammasome-mediated IL-1b maturation in a dose-dependent manner. THP-1 lysates from LPS-prestimulated cells were incubated at 30 8C in the presence of either GST (control) or GST-rat caspase-12, and IL-1b maturation was detected by western blot analysis using an antibody against the mature form of IL-1b. c, Caspase-12 co-immunoprecipitates with caspase-1. HEK 293T cells were co-transfected with Flag-tagged caspase-1, caspase-5 or caspase-9 and caspase-12. Flag-tagged caspases were immunoprecipitated from HEK 293T cells with Flag M2 agarose beads and the association of caspase-12 was examined by western blot of the immunoprecipitated product using antimouse caspase-12 antibodies. ‘Total’ represents 10% of the total cell lysates used in the immunoprecipitation (IP) reaction. The caspase-12 processing products probably result from autocatalysis as they were not present when the catalytically inactive caspase-12 C299A mutant was transfected in 293T cells (not shown).

positioned in the caecal wall. After surgery, mice were monitored five times per day during the first 48 h after surgery, observing activity, alertness and temperature. Animals that were determined to be moribund (temperature ,23 8C, showing minimal response to challenge and diminished righting reflex) were killed, allowing for collection of blood and for post-mortem analysis. Long-term survivors were followed until day 8 after surgery when they were killed for final sampling and necropsy. Listeria monocytogenes challenge. For in vivo infections, Listeria monocytogenes strain DP-L4056 (a gift from D. Portnoy) was grown to mid-logarithmic phase. Approximately 5 £ 104 c.f.u. were injected into the lateral tail vein. Determination of bacterial content in sepsis and Listeria models. In the sepsis model, 100 ml peripheral blood samples were collected by orbital puncture 12 h after CASP. The blood volume was brought up to 0.5 ml with distilled water and serial dilutions of that (1:2, 1:100, 1:1,000) were prepared in distilled water and were plated on brain heart infusion blood agar plates. After a 24-h incubation of the plates, bacterial colonies were counted and the number of c.f.u. per ml blood was calculated. In the Listeria model, peripheral blood was collected and analysed as above on days 1, 2, 3 and 4 after Listeria injection. At the same time points, spleens and livers were also collected and were homogenized in a 10-ml volume of a 0.2% NP-40 solution. Serial dilutions (1:10, 1:100, 1:1,000) were plated on brain heart infusion agar and bacterial counts were assessed as above. Splenocyte culture and treatments. Splenocytes were isolated by mechanical disruption followed by differential centrifugation and re-suspension in RPMI.

© 2006 Nature Publishing Group

1067

LETTERS

NATURE|Vol 440|20 April 2006

Cells were treated for 24 h with PAMcsk4 (100 ng ml21, TLR2), poly(I:C) (4 mg ml21, TLR3), LPS Re595 (Salmonella minnesota; 100 ng ml21, TLR4), resiquimod R848 (1:500, TLR7), CpG (0.2 mM, TLR9), and muramyl dipeptide MDP (10 mM or 100 mM) with or without cycloheximide CHX (0.5 mg ml21). Cells were then treated with 5 mM ATP in fresh media for 20 min, washed and further cultured in fresh media for 3 h. A 10–plex (ten mouse cytokines: IL-1b, IL-2, IL-4, IL-5, IL-6, Il-10, IL-12p70, IFN-g, TNF-a and GM-CSF) bead-based immunoassay kit from LINCOplex was used according to the manufacture’s instructions. Data were obtained and analysed using the Bio-Plex system from Biorad. Anti-IFN-g receptor treatment. Anti-IFN-g antibodies and the isotype control antibodies were purchased from PharMingen (catalogue number 557530 and 553969, respectively). The antibodies were administered via intraperitoneal injection at a dose of 250 mg per mouse 1 h before induction of sepsis by CASP. THP-1 cell transfection and differentiation. THP-1 cells were transfected by electroporation with 20 mg DNA (15 mg caspase-12 (either wild type or Cys299Ala mutant) plus 5 mg GFP). After electroporation, cells were kept on ice for 5 min then grown in 20 ml media for 24 h. Four electroporation reactions (that is, 80 million cells) were pooled for each caspase-12 (either wild type or Cys299Ala mutant) transfection. The next day, live GFP-positive cells were sorted, pre-treated with 100 ng ml21 phorbol myristate acetate (PMA) for 6 h then treated with TLR ligands for 24 h. For IL-1b measurement, cells were then treated with 5 mM ATP in fresh media for 20 min, washed and further cultured in fresh media for 3 h. Cell supernatants were then analysed for cytokine production using bead-based multiplex human immunoassays (described above). Caspase-1 activity assays. HEK 293T cells were co-transfected using lipofectamine plus reagent (Invitrogen) with Flag-tagged caspase-1 and caspase-12 (either wild type or Cys299Ala mutant). Twenty-four hours after transfection, cell lysates were prepared in caspase activation buffer (20 mM PIPES pH 7.4, 100 mM NaCl, 1 mM EDTA, 10 mM dithiothreitol (added fresh), 0.1 CHAPS, 10% sucrose and protease inhibitors), and 100 mg lysates were assayed for cleavage of 100 mM of the caspase-1 substrate WEHD-AFC. Kinetic measurements of AFC release were recorded every minute for 30 min at room temperature using a fluorescent plate reader (excitation 400 nm, emission 505 nm). Cell-free inflammasome activation assay. As in ref. 11, THP-1 cells were prestimulated with 1 mg ml21 LPS for 1 h. Cells were harvested and washed with PBS, then lysed by hypo-osmotic swelling followed by mechanical disruption by passage 15 times through a 22-gauge needle. Cell lysates were centrifuged, and the supernatants, after filtration (0.45 mM), were used for this in vitro IL-1b cleavage assay. The lysates were incubated for 20 min on ice with either GST (control) or GST-rat caspase-12, then shifted to 30 8C and incubated for 60 min. IL-1b processing was monitored by western analysis with a specific antibody against the mature form of IL-1b (Cell Signaling, catalogue number 2021). Co-immunoprecipitation experiments. HEK 293T cells were co-transfected using lipofectamine plus reagent (Invitrogen) with Flag-tagged caspase-1, caspase-5, caspase-9 (Cys-to-Ala mutants) and caspase-12. Twenty-four hours after transfection, cell lysates were obtained by detergent disruption and caspase1, caspase-5 and caspase-9 were immunoharvested using Flag M2 agarose beads (Sigma). Immunocomplexes were eluted from the beads using Flag peptides (Sigma) and were processed for western blot analysis using anti-mouse caspase-12 antibodies or anti-Flag antibodies (Sigma). Other methods. Cell death was triggered in MEFs and was measured by flow cytometry after propidium iodide staining. Immunohistochemical procedures were performed on sections fixed in 1% formaldehyde. Bone marrow chimaeric mice were generated after transfer of harvested bone marrow into recipient mice that had been ablated with two doses of irradiation at 550 rad. NF-kB activity was determined in cells that had been transfected with kB-luciferase and b-gal reporter plasmids. Detailed descriptions of these and all methods are available in Supplementary Information. Statistical analysis. Log rank test and Fisher’s exact test were used to compare sepsis survival among Casp12 þ/þ and Casp12 2/2 mice. Student t-test was used

1068

for statistical analyses of apoptosis, bacterial clearance as well as cytokine production assays. Received 6 December 2005; accepted 14 February 2006. 1. 2.

3. 4. 5. 6.

7.

8.

9.

10. 11.

12.

13.

14. 15. 16. 17.

18.

Hotchkiss, R. S. et al. Caspase inhibitors improve survival in sepsis: a critical role of the lymphocyte. Nature Immunol. 1, 496–-501 (2000). Li, P. et al. Mice deficient in IL-1b-converting enzyme are defective in production of mature IL-1b and resistant to endotoxic shock. Cell 80, 401–-411 (1995). Kuida, K. et al. Altered cytokine export and apoptosis in mice deficient in interlukin-1b converting enzyme. Science 267, 2000–-2003 (1995). Wang, S. et al. Murine caspase-11, an ICE-interacting protease, is essential for the activation of ICE. Cell 92, 501–-509 (1998). Saleh, M. et al. Differential modulation of endotoxin responsiveness by human caspase-12 polymorphisms. Nature 429, 75–-79 (2004). Zantl, N. et al. Essential role of gamma interferon in survival of colon ascendens stent peritonitis, a novel murine model of abdominal sepsis. Infect. Immun. 66, 2300–-2309 (1998). Hotchkiss, R. S. et al. Adoptive transfer of apoptotic splenocytes worsens survival, whereas adoptive transfer of necrotic splenocytes improves survival in sepsis. Proc. Natl Acad. Sci. USA 100, 6724–-6729 (2003). Prass, K. et al. Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke T helper cell type 1-like immunostimulation. J. Exp. Med. 198, 725–-736 (2003). Rasper, D. M. et al. Cell death attenuation by ‘Usurpin’, a mammalian DED-caspase homologue that precludes caspase-8 recruitment and activation by the CD-95 (Fas, APO-1) receptor complex. Cell Death Differ. 5, 271–-288 (1998). Irmler, M. et al. Inhibition of death receptor signals by cellular FLIP. Nature 388, 190–-195 (1997). Martinon, F., Burns, K. & Tschopp, J. The inflammasome: a molecular platform triggering activation of inflammatory caspases and processing of proIL-b. Mol. Cell 10, 417–-426 (2002). Joshi, V. D., Kalvakolanu, D. V., Hebel, J. R., Hasday, J. D. & Cross, A. S. Role of caspase 1 in murine antibacterial host defenses and lethal endotoxemia. Infect. Immun. 70, 6896–-6903 (2002). Watanabe, T., Kitani, A., Murray, P. J. & Strober, W. NOD2 is a negative regulator of Toll-like receptor 2-mediated T helper type 1 responses. Nature Immunol. 5, 800–-808 (2004). Kobayashi, K. S. et al. Nod2-dependent regulation of innate and adaptive immunity in the intestinal tract. Science 307, 731–-734 (2005). Nakagawa, T. et al. Caspase-12 mediates endoplasmic reticulum-specific apoptosis and cytotoxicity by amyloid-b. Nature 403, 98–-103 (2000). Kalai, M. et al. Regulation of the expression and processing of caspase-12. J. Cell Biol. 162, 457–-467 (2003). Obeng, E. A. & Boise, L. H. Caspase-12 and caspase-4 are not required for caspase-dependent endoplasmic reticulum stress-induced apoptosis. J. Biol. Chem. 280, 29578–-29587 (2005). Hitomi, J. et al. Involvement of caspase-4 in endoplasmic reticulum stressinduced apoptosis and Ab-induced cell death. J. Cell Biol. 165, 347–-356 (2004).

Supplementary Information is linked to the online version of the paper at www.nature.com/nature. Acknowledgements M.S. is supported by a CIHR post-doctoral fellowship. D.R.G. is supported by grants from the US NIH. We thank S. Granger and A. Coddington for help with the Bio-plex system and C. Bonzon for help with MEF preparation. Author Contributions D.R.G. and D.W.N. share senior authorship. Author Information Reprints and permissions information is available at npg.nature.com/reprintsandpermissions. The authors declare competing financial interests: details accompany the paper on www.nature.com. Correspondence and requests for materials should be addressed to D.W.N ([email protected]) or D.R.G. ([email protected]).

© 2006 Nature Publishing Group